Structure of the methanol synthesis catalyst determined by in situHERFD XAS and EXAFS

Abstract

A Cu/ZnO/Al₂O₃ commercial catalyst for methanol synthesis from syngas was investigated under operational conditions. HERFD XAS and EXAFS data were recorded under different reaction gas mixtures, temperatures, and pressures. Activation of the catalyst precursor occurred via a Cu⁺ intermediate. The active catalyst predominantly consists of metallic Cu and ZnO. Methanol production only starts when all accessible Cu is reduced. The structure of the active catalyst did not change with temperature or pressure even though the methanol yield changed strongly. Formation of a carbon-containing layer on top of the catalyst surface was detected by TPD, which was correlated with a several-hour induction period in the methanol production after the catalyst reduction.

---

1. Introduction

Methanol synthesis is a process of major importance for the chemical industry. Apart from its common use as solvent, methanol is used in the production of other chemicals, mainly formaldehyde, which is a raw material for producing various types of plastics.¹Methanol is synthesized via a process implemented by Imperial Chemical Industries in 1966. Synthesis gas (syngas), which is a mixture of carbon monoxide, hydrogen and carbon dioxide, is converted to methanol over a Cu/ZnO/Al₂O₃ catalyst, between 50 and 100 bar and at 250 °C, with selectivities approaching 100%.¹

The mechanism of methanol synthesis is yet to be understood, especially the nature of the active sites. It has been suggested that methanol formation is catalyzed by small amounts of non-metallic Cu^2,3 or by negatively charged Cu species at the so-called Schottky junction between Cu metal and semiconducting ZnO.⁴ However, the common agreement is that the active catalyst consists of metallic Cu and ZnO.^1,5–11 An alloy between Cu and Zn^12–15 and ZnO on top of metallic Cu^16,17 have also been suggested as possible active species.

Two preferred methods to characterize the catalyst structure under working conditions are synchrotron-based X-ray absorption spectroscopy (XAS) and X-ray diffraction (XRD). In situXAS showed that Cu is metallic when supported on ZnO and SiO₂.¹⁸ Previous studies demonstrated that Cu particles supported on ZnO respond differently compared to the SiO₂ supported particles to changes in the reduction potential of the syngas.¹⁹ Whereas Cu/SiO₂ underwent no changes, Cu/ZnO showed significant morphological changes, which was explained by changes in the relative surface and interface energies. Grunwaldt et al.,¹⁸ by means of combined XAS, XRD, and online catalytic measurements, concluded that Cu particles change shape reversibly depending on the reaction conditions, which suggested the wetting of ZnO by Cu.

The XAS investigations referred to above focused mainly on the extended X-ray absorption structure (EXAFS) region of the spectrum, which reflects the local geometry of the compound, while the chemical state of the metal atom can be obtained from the X-ray absorption near-edge structure (XANES) region. Analysis of the XANES region is complicated by the core–hole lifetime broadening, which smears out features in the spectrum and makes distinguishing between chemical states difficult. In 1976, Eisenberger et al.²⁰ showed that the problem of the lifetime broadening in the XANES region could be overcome by detecting the X-ray absorption spectrum using the intensity of the emitted X-ray fluorescence in a narrow energy bandwidth. Later, the technique was called high-energy-resolution fluorescence detected (HERFD) XAS.²¹ The benefit of recording a high resolution XAS spectrum becomes obvious if considering pre-edges of the third-row transition metals, which arise from quadrupole transitions to unoccupied 3d states.²² These small peaks located a few eV below the K-edge become clearly resolved and serve as a direct probe of the chemical state. In particular, it allows Cu⁺ and Cu²⁺ to be distinguished unambiguously.^23,24 With recent instrumentation development,^22,25–36 few synchrotrons made the HERFD XAS technique available to the XAS community. Applications of HERFD XAS to catalysis include the study of Fe/ZSM-5 by Heijboer et al.,³⁷Mn/ZSM-5 by Radu et al.,³⁸Au/Al₂O₃ by van Bokhoven et al.,³⁹ and Pt/Al₂O₃ by Singh et al.⁴⁰

The aim of the experiment reported is to evaluate the state of the Cu/ZnO/Al₂O₃ catalyst under reaction conditions, which is achieved with operando HERFD XAS at realistic temperature, pressure, and reaction gas mixture composition. EXAFS was taken simultaneously with HERFD XAS to obtain the local geometric structure and correlate the results with those of similar EXAFS experiments performed by other groups.^8,18,41,42

2. Experimental

The Cu/ZnO/Al₂O₃ catalyst was prepared by a conventional co-precipitation method with a wt% ratio of CuO : ZnO : Al₂O₃ = 6 : 3 : 1.^1,43–45Catalytic tests under industry relevant conditions were carried out with 250 mg of catalyst in a stainless steel reactor tube with a diameter of 6 mm. Prior to the reaction the catalyst was reduced for 6 hours in a stream of 5% H₂/He at 250 °C with a total flow rate of 75 ml min⁻¹. The reaction feed composition was 31% CO, 4% CO₂, 3% Ar, H₂ balance, with a total flow rate of 10 ml min⁻¹. The catalytic tests were performed at 200–250 °C and 20–50 bar.

In situ XAS experiments were performed with a setup shown in Fig. 1. The reactor is a glass capillary with an outer diameter of 3.0 mm, and a wall thickness of 0.1 mm. The powder catalyst was fixed in the reactor by two quartz wool plugs. The temperature of the catalyst was regulated with an air blower (FMB Oxford) placed below the sample. Pressure in the reactor was adjusted with a leak valve at the exit and monitored with a mechanical pressure gauge. The gas flow was regulated with mass-flow controllers. The maximum pressure in the reactor, which can be achieved with the setup, is roughly 7 bar. Reaction products were monitored with a quadrupole mass-spectrometer (Pfiffer Ministar).

Fig. 1 Scheme of the in situ experiment. MFC: mass-flow controller, P: pressure meter, QMS: quadrupole mass spectrometer.

All the in situ experiments were performed with a total gas flow between 20 and 30 ml min⁻¹. Approximately 20 mg of catalyst was used in every run. The catalyst was treated as follows: the precursor was activated in situ by reducing it in 5% H₂/He or reaction mixture (2.5% CO, 0.3% CO₂, and 5% H₂ in He) at 250 °C. After reduction in H₂/He, the catalyst was cooled down to room temperature, the gas changed to the reaction mixture, and the catalyst heated to 250 °C. The heating rate did not exceed 5 °C min⁻¹.

In situ XAS experiments were performed at the SuperXAS beamline of the Swiss Light Source. The beamline provides an X-ray flux of 6 × 10¹¹photon s⁻¹ and an energy bandwidth of 1 eV at the Cu K-edge. The spot size at the sample was 0.1 mm (vertical) by 0.2 mm (horizontal). Emitted X-ray radiation was registered with a Johann-type spectrometer²⁵ with five spherically-curved Si(111) crystals and a PILATUS 100 K detector. A bag with kapton windows filled with helium was placed between the sample, crystals, and detector to minimize absorption of X-rays by air. The energy bandwidth of the spectrometer at the Cu Kα1 emission line was below 1 eV. The reactor diameter was too large to record X-ray transmission; therefore Cu K-edge XAS spectra were detected with fluorescence using a high-resolution X-ray emission spectrometer, and the Zn K-edge spectra were detected with an energy-dispersive silicon-drift detector (KETEK).

XRD (X-ray diffraction) was performed in a Philips X'Pert, MPD = DY636, CuKα, λ = 1.540598 Å, 2θ = 20–80°, step size 0.05°, using ‘Xpert Highscore’ Software.

3. Results and discussion

Table 1 shows the performance of the catalyst at 20 and 50 bar. The conversion of CO and methanol yield increase with pressure and temperature up to 250 °C, which was found to be optimal for the methanol synthesis.

Table 1 Performance of the catalyst under industry-relevant conditions

P/bar	T/°C	CO conversion/%	Methanol yield/%
20	200	13	4.6
20	250	20	7.7
50	200	17	6.0
50	250	40	16.0

---

The size of ZnO and CuO crystallites in the catalyst precursor after calcination was 7 nm, as estimated from ZnO(100) and CuO(111) XRD peak widths. Assuming the thickness of the surface of the topmost layer of CuO and ZnO to be 2.5 Å, the particle of that size contains about 10% atoms in the surface layer, i.e.XAS, which is bulk sensitive, is able to detect surface changes during reaction.

Fig. 2 reports Cu K-edge HERFD XAS spectra of the calcined catalyst and that of the catalyst reduced in 5% hydrogen at 250 °C for 1 h. Spectra of Cu, Cu₂O, and CuO are shown for comparison. The Cu K-edge originates from the transition of a Cu 1s electron to an unoccupied 4p state.

Fig. 2 Cu K-edge HERFD XAS of the catalyst before (blue) and after (red) reduction in hydrogen and spectra of reference Cu compounds (dashed).

As the transition energy depends on the ion charge, the edge position shifts towards lower energies with oxidation states: 8979.7 eV, 8981.1 eV, and 8984.3 eV for Cu, Cu₂O, and CuO respectively. The spectral region above the edge, the so-called ‘multiple scattering’ region is formed by transitions to the continuum of states above the vacuum level. Although, in general individual spectral features cannot be assigned unambiguously, the spectral region serves as a fingerprint of the structure, ligand type, and chemical state of the atom. Small shifts in the white-line with respect to the references can be due to difference in particle size of the catalyst and references. Therefore we presented the full HERFD spectrum so comparison is not done merely on the white-line position but on the overall spectrum shape. The pre-edge, a small peak in the spectrum of CuO at 8979 eV, originates from the weak quadrupole transition of the 1s electron to an unoccupied 3d state. Cu²⁺ ions have one unfilled 3d orbital, while all 3d orbitals are filled in Cu⁺ and Cu⁰. Therefore, this pre-edge serves as a clear indication of Cu²⁺.

The spectrum of the calcined catalyst (Fig. 2) is similar to that of CuO, although the pre-edge is shifted towards higher energy and features are slightly broader, which can be accounted for by the disorder in the structure of nanoparticles. The quantification of the relative concentration of atoms in different chemical states is performed with linear combination fitting (LCF). The accuracy of LCF depends largely on the reference spectra used. To evaluate how many spectral components are required to reproduce the spectrum, we applied principal component analysis (PCA)^46,47 to the set of spectra obtained during reduction. PCA is a mathematical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of uncorrelated variables called principal components.⁴⁸ It is a standard procedure used in chemometrics to reduce data dimension. Analysis was performed in such a way that our principal components justify ∼95% of data variability. LCF was performed using references as close to the system used since this is the major source of errors in the fitting.

We found that three spectral components are required, which means that three oxidation states of Cu are present. We used a spectrum of the catalyst at 450 °C in oxygen and at 250 °C in hydrogen as references for Cu²⁺ and Cu⁰. As no reliable spectrum of the catalyst in a possible Cu⁺ state was available, we used a spectrum of Cu₂O broadened by convolution with a Gaussian and checked with the PCA transformation.⁴⁶LCF reproduced the spectra with an accuracy of 5% or better, which is a good estimate for the uncertainty in the concentrations obtained.

Fig. 3a shows the increase in Cu²⁺, Cu⁺, and Cu⁰ concentrations with temperature during reduction in the reaction mixture. The concentrations are correlated with CO₂ and H₂O MS signals in Fig. 3b. Three stages can be identified in the catalyst reduction. Below 50 °C, the spectra showed no change in spectral features and reactants conversion. The spectra were fully fitted with that of the Cu²⁺ reference. The next stage ranged from 50 to 175 °C, where Cu⁺ and Cu⁰ phases are detected, resulting from the reduction of Cu²⁺ due to CO and H₂. Fig. 3b shows that CO and H₂O were formed during the reduction suggesting that both CO and H₂ reduce the precursor. The final stage was characterized by an almost complete depletion of CuO and the presence of Cu metallic phase.

Fig. 3 (a) Concentration of Cu⁰ (black), Cu⁺ (red), and Cu²⁺ (blue) derived from the linear combination fit of Cu K-edge HERFD XAS. (b) Water (solid black) and carbon dioxide (red dashed) formation during reduction in the reaction mixture at 6 bar, with a heating ramp of 2 °C min⁻¹.

Fig. 4a shows the evolution of the methanol signal during and after reduction in the reaction mixture and continued exposure to the reaction mixture at 250 °C and 6 bar. The first 2 hours in Fig. 4a are same as those in Fig. 3 while heating. More than 90% of Cu atoms got reduced by reaching a temperature of 250 °C. The lingering 10% of the total copper content remained oxidized. Since TPR experiments suggested full reduction below 200 °C we assume that this amount is associated with some adsorbed species that are difficult to remove, such as formates. Due to their stability and long-lived signal we believe that these species act as spectators and play no role in the catalysis, except to decrease the amount of available copper sites for reaction.

Fig. 4 (a) Temperature (blue dashed) and methanol yield (solid black) during treatment of the catalyst precursor in the reaction mixture. (b) Cu K-edge HERFD XAS recorded at the time points marked in (a).

Methanol formation only starts when no further reduction occurred (point A). Thus, all the accessible and reducible Cu²⁺ is reduced prior to methanol formation. After a continuous rise, which lasts for about 4 hours, the methanol signal growth slows down (point B) and reaches a plateau approximately 10 hours after the reduction. After the signal reached a steady state, the reactor was cooled down to room temperature (point C), which caused the methanol signal to drop to the initial level. The average conversion at 6 bar and 250 °C is around 5%.

Fig. 4b shows the Cu K-edge HERFD XAS taken at time points A, B, and C shown in Fig. 4a. No change in the spectrum shape can be detected within the accuracy of the measurements, which suggest that the increase in the methanol yield is not correlated with any change in the oxidation state of Cu and should be caused by some other process.

Following cooling down to room temperature, a temperature programmed desorption in argon was performed (ESI†). The temperature was increased stepwise to 500 °C. At every increase in temperature, a massive desorption of carbon-containing species was detected.

Fig. 5a shows the Fourier-transformed Cu K-edge EXAFS of the catalyst under the reaction feed at three different temperatures. Heating in hydrogen prior to the EXAFS experiment activated the catalyst. The shape of the Fourier transformed spectrum does not change significantly while heating from 150 °C to 250 °C or while cooling down to 20 °C, except for a noticeable variation in the amplitude. Zn K-edge EXAFS taken under the same conditions also show only fluctuations in the amplitude and constant shape (Fig. 5b).

Fig. 5 Fourier-transformed (a) Cu K-edge and (b) Zn K-edge EXAFS of the catalyst in the reaction mixture at 150 °C (solid black), 250 °C (red dashed), and 20 °C (blue dot-dashed).

Fourier-transformed EXAFS spectra were fit with a one-shell Cu metal and ZnO structures using Artemis,⁴⁹ IFEFFIT,⁵⁰ and FEFF⁵¹ software packages. Table 2 shows the parameters of the fit. The energy shift ΔE was fixed at a value found in a preliminary run. Neither the Cu–Cu nor the Zn–O bond length experiences any change with temperature. Relative changes in the coordination number are within the experimental uncertainty. The Debye–Waller disorder parameter σ of the Cu–Cu bond increases with temperature, which explains the variation in EXAFS amplitude as shown in Fig. 5a. The disorder parameter of the Zn–O bond did not change. The Cu K-edge EXAFS can be equally well fitted with a structure where an arbitrary number of Cu atoms were exchanged with Zn. This is because Cu and Zn atoms are neighbors in the chemical table and their EXAFS scattering phases are very similar.

Table 2 EXAFS fit parameters. N: coordination number, N₀: arbitrary constant, σ: disorder parameter, R: internuclear distance, ΔE₀: energy shift. ΔE₀ was fixed at values obtained in the preliminary fit

		150 °C	250 °C	20 °C
Cu–Cu	N/N0	1.0 ± 0.1	1.0 ± 0.2	1.1 ± 0.1
	σ ^2/Å^2	0.011 ± 0.002	0.014 ± 0.002	0.009 ± 0.001
	R/Å	2.54 ± 0.01	2.54 ± 0.01	2.54 ± 0.01
	ΔE0/eV	3.8	3.8	3.8
Zn–O	N/N0	1.0 ± 0.1	1.0 ± 0.1	1.0 ± 0.1
	σ ^2/Å^2	0.004 ± 0.002	0.005 ± 0.002	0.004 ± 0.001
	R/Å	1.97 ± 0.01	1.97 ± 0.01	1.97 ± 0.01
	ΔE0/eV	4.7	4.7	4.7

---

Fig. 6 depicts methanol formation evaluated as a function of pressure of the reaction gas mixture after reduction in hydrogen. The methanol signal increased with pressure and temperature. As expected, increasing pressure favored methanol formation (Table 1). The increase in pressure did not cause any detectable change in the oxidation state and/or structure (HERFD XAS and EXAFS spectra not shown here). The values of methanol production are represented in a linear scale and the maximum value of methanol produced relates to 5% conversion.

Fig. 6 Methanol formation (solid black) dependence on pressure (blue dashed) and temperature (red dot-dashed).

The results of the in situ experiment correlate well with XAS data reported by other groups. Reduction of the Cu/Al₂O₃ catalyst observed by Stötzelet al.⁴² with QXAFS with a sub-second time resolution followed the same stepwise Cu⁰ → Cu⁺ → Cu²⁺ mechanism, as we detected with HERFD XAS (Fig. 3a). The CO₂ signal, originating from the oxidation of CO, peaked at lower temperature than the signal of H₂O, 140 °C compared to 160 °C (Fig. 3b), which is slightly lower than that found by Grunwaldt et al.¹⁸ for a supported Cu/ZnO catalyst. The results suggest that copper reduction under hydrogen or reaction gas mixture atmospheres occurred below 250 °C. Zn K-edge HERFD XAS and EXAFS showed no significant change during the reduction (ESI†).

Under reaction conditions, the major Cu structure is that of the Cu metal, and zinc is in a ZnO phase as the analysis of EXAFS and HERFD XAS suggests. The structure of the active catalyst is achieved upon reduction. Under the reaction gas mixture the structure of the reduced catalyst is not sensitive to temperature or pressure even though the yield of methanol strongly varies. The XAS data obtained neither support nor disprove the hypothesis of CuZn alloying at the surface of the active catalyst.^12,18Cu K-edge EXAFS fits equally well the Cu metal structure, or the structure where an arbitrary number of Cu atoms is replaced by Zn. HERFD XAS of a CuZn alloy (brass) differ from that of the metallic Cu (ESI†), but this may account for the differences in the internuclear distances. EXAFS data show that the Cu–Cu or Cu–Zn distance remains constant and matches perfectly with the Cu–Cu distance in the Cu metal. All these allow us to conclude that if a Cu–Zn alloy was formed, it was below the detection limit of our method.

The rather large period measured before achieving stable methanol production (Fig. 4) suggests that some of the methanol and/or reaction intermediates were used in subsequent reactions that form larger organic molecules. The steady increase in the methanol yield indicates that the carbon-containing surface layer does not completely poison the active sites, but rather contains intermediates of the reaction, which correlates with the results of IR studies of the Cu/ZnO catalyst reported by Edwards and Schrader⁵² and chemical trapping by Kieffer et al.,⁵³ who detected various formates as intermediates in the methanol formation reaction.

4. Conclusions

We characterized the electronic and geometric structure of a Cu/ZnO/Al₂O₃ catalyst during pretreatment and during methanol formation at 6 bar. We identified the intermediate Cu⁺ state during reduction of the catalyst. The working catalyst predominantly consists of Cu⁰, with a structure of Cu metal, and ZnO. Methanol production only starts when all accessible Cu is reduced. The structure of the reduced catalyst was not changing with pressure, temperature, and time on stream even though the yield of methanol strongly increased.

A few hour induction period in methanol production was observed after the reduction of the catalyst, caused by the formation of a carbon-containing layer on top of the catalyst surface, which presumably contains intermediates of the methanol formation reaction.

Acknowledgements

We thank Erich de Boni (PSI) for help in development of the in situ catalytic setup and Alwin Frei (PSI) for XRD data. Evgeny Kleymenov thanks the Swiss National Science Foundation for financial support (grant 200021-119881). We thank the Swiss Light Source for providing the beam time.

References

1. G. Chinchen, P. Denny, J. Jennings, M. Spencer and K. Waugh, Appl. Catal., 1988, 36, 1–65.
2. K. Klier, Adv. Catal., 1982, 31, 243–313.
3. V. Ponec, Surf. Sci., 1992, 272, 111–117.
4. J. C. Frost, Nature, 1988, 334, 577–580.
5. T. H. Fleisch and R. L. Mieville, J. Catal., 1984, 90, 165–172.
6. M. Muhler, L. P. Nielsen, E. Tornqvist, B. S. Clausen and H. Topsoe, Catal. Lett., 1992, 14, 241–249.
7. K. Waugh, Catal. Today, 1992, 15, 51–75.
8. B. S. Clausen and H. Topsoe, Catal. Today, 1991, 9, 189–196.
9. P. Rasmussen, M. Kazuta and I. Chorkendorff, Surf. Sci., 1994, 318, 267–280.
10. P. B. Rasmussen, P. M. Holmblad, T. Askgaard, C. V. Ovesen, P. Stöltze, J. K. Nrskov and I. Chorkendorff, Catal. Lett., 1994, 26, 373–381.
11. J. Yoshihara and C. T. Campbell, J. Catal., 1996, 161, 776–782.
12. N.-Y. Topsoe and H. Topsoe, Top. Catal., 1999, 8, 267–270.
13. T. Fujitani and J. Nakamura, Catal. Lett., 1998, 56, 119–124.
14. T. Fujitani, I. Nakamura, T. Uchijima and J. Nakamura, Surf. Sci., 1997, 383, 285–298.
15. N.-Y. Topsoe and H. Topsoe, J. Mol. Catal. A: Chem., 1999, 141, 95–105.
16. M. M. Viitanen, W. P. A. Jansen, R. G. van Welzenis, H. H. Brongersma, D. S. Brands, E. K. Poels and A. Bliek, J. Phys. Chem. B, 1999, 103, 6025–6029.
17. E. K. Poels and D. S. Brands, Appl. Catal., A, 2000, 191, 83–96.
18. J. D. Grunwaldt, A. M. Molenbroek, N. Y. Topse, H. Topsoe and B. S. Clausen, J. Catal., 2000, 194, 452–460.
19. B. S. Clausen, J. Schiotz, L. Grabk, C. V. Ovesen, K. W. Jacobsen, J. K. Norskov and H. Topsoe, Top. Catal., 1994, 1, 367–376.
20. P. Eisenberger, P. M. Platzman and H. Winick, Phys. Rev. Lett., 1976, 36, 623–626.
21. O. V. Safonova, M. Tromp, J. A. van Bokhoven, F. M. F. de Groot, J. Evans and P. Glatzel, J. Phys. Chem. B, 2006, 110, 16162–16164.
22. P. Glatzel and U. Bergmann, Coord. Chem. Rev., 2005, 249, 65–95.
23. L. S. Kau, D. J. Spira-Solomon, J. E. Penner-Hahn, K. O. Hodgson and E. I. Solomon, J. Am. Chem. Soc., 1987, 109, 6433–6442.
24. E. Kleimenov, E. Alayon, J. van Bokhoven, A. Eliseev, M. Janousch, N. Verbitskij, A. Vinogradov and M. Nachtegaal, Chimia, 2010, 64, 572.
25. E. Kleymenov, J. A. van Bokhoven, C. David, P. Glatzel, M. Janousch, R. Alonso-Mori, M. Studer, M. Willimann, A. Bergamaschi, B. Henrich and M. Nachtegaal, Rev. Sci. Instrum., 2011, 82, 065107.
26. V. Stojanoff, K. Hamalainen, D. Siddons, J. Hastings, L. Berman, S. Cramer and G. Smith, Rev. Sci. Instrum., 1992, 63, 1125–1127.
27. W. Schulke, A. Kaprolat, T. Fischer, K. Hoppner and F. Wohlert, Rev. Sci. Instrum., 1995, 66, 2446–2452.
28. U. Bergmann and S. Cramer, Proc. SPIE–Int. Soc. Opt. Eng., 1998, 3448, 198–209.
29. U. Bergmann, P. Glatzel and S. P. Cramer, Microchem. J., 2002, 71, 221–230.
30. H. Hayashi, M. Kawata, R. Takeda, Y. Udagawa, Y. Watanabe, T. Takano, S. Nanao and N. Kawamura, J. Electron Spectrosc. Relat. Phenom., 2004, 136, 191–197.
31. G. Tirao, G. Stutz and C. Cusatis, J. Synchrotron Radiat., 2004, 11, 335–342.
32. B. Dickinson, G. T. Seidler, Z. W. Webb, J. A. Bradley, K. P. Nagle, S. M. Heald, R. A. Gordon and I. M. Chou, Rev. Sci. Instrum., 2008, 79, 123112.
33. E. Kleimenov, A. Bergamaschi, J. van Bokhoven, M. Janousch, B. Schmitt and M. Nachtegaal, J. Phys.: Conf. Ser., 2009, 190, 012035.
34. L. Journel, L. E. Khoury, T. Marin, R. Guillemin, S. Carniato, A. Avila, R. Delaunay, C. F. Hague and M. Simon, Rev. Sci. Instrum., 2009, 80, 093105.
35. J.-L. Hazemann, O. Proux, V. Nassif, H. Palancher, E. Lahera, C. Da Silva, A. Braillard, D. Testemale, M.-A. Diot, I. Alliot, W. Del Net, A. Manceau, F. Gelebart, M. Morand, Q. Dermigny and A. Shukla, J. Synchrotron Radiat., 2009, 16, 283–292.
36. R. Verbeni, T. Pylkkanen, S. Huotari, L. Simonelli, G. Vanko, K. Martel, C. Henriquet and G. Monaco, J. Synchrotron Radiat., 2009, 16, 469–476.
37. W. Heijboer, D. Koningsberger, B. Weckhuysen and F. de Groot, Catal. Today, 2005, 110, 228–238.
38. D. Radu, P. Glatzel, A. Gloter, O. Stephan, B. M. Weckhuysen and F. M. F. de Groot, J. Phys. Chem. C, 2008, 112, 12409–12416.
39. J. A. van Bokhoven, C. Louis, J. T. Miller, M. Tromp, O. V. Safonova and P. Glatzel, Angew. Chem., Int. Ed., 2006, 45, 4651–4654.
40. J. Singh, E. M. C. Alayon, M. Tromp, O. V. Safonova, P. Glatzel, M. Nachtegaal, R. Frahm and J. A. van Bokhoven, Angew. Chem., Int. Ed., 2008, 47, 9260–9264.
41. B. S. Clausen, B. Lengeler and B. S. Rasmussen, J. Phys. Chem., 1985, 89, 2319–2324.
42. J. Stötzel, D. Lutzenkirchen-Hecht, R. Frahm, B. Kimmerle, A. Baiker, M. Nachtegaal, M. J. Beier and J.-D. Grunwaldt, J. Phys.: Conf. Ser., 2009, 190, 012153.
43. E. Doesburg, R. Hoppener, B. de Koning, X. Xiaoding and J. Scholten, Stud. Surf. Sci. Catal., 1987, 31, 767–783.
44. R. Hoppener, E. Doesburg and J. Scholten, Appl. Catal., 1986, 25, 109–119.
45. L. I. Kuznetsova, T. M. Yurieva, T. P. Minyukova, S. V. Ketchik, L. M. Plyasova and G. K. Boreskov, React. Kinet. Catal. Lett., 1982, 19, 355–359.
46. S. R. Wasserman, J. Phys. IV, 1997, 7, C2-203–C2-205.
47. E. Malinowski, Factor analysis in chemistry, Wiley-Interscience, 3rd edn, 2002.
48. K. Pearson, Philos. Mag., 1901, 2, 559–572.
49. B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12, 537–541.
50. M. Newville, J. Synchrotron Radiat., 2001, 8, 322–324.
51. J. J. Rehr and R. C. Albers, Rev. Mod. Phys., 2000, 72, 621–654.
52. J. F. Edwards and G. L. Schrader, J. Phys. Chem., 1984, 88, 5620–5624.
53. R. Kieffer, E. Ramaroson, A. Deluzarche and Y. Trambouze, React. Kinet. Catal. Lett., 1981, 16, 207–212.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c1cy00277e

---